Torchvision - can't extract Caltech256

[ikamensh]

🐛 Bug

When I try to download Caltech256 dataset via torchvision, an empty file is downloaded.

-rw-r--r-- 1 ikkamens staff 0B Jun 26 12:05 256_ObjectCategories.tar

I get following error:

0it [00:00, ?it/s]
Traceback (most recent call last):
  File "/Users/ikkamens/PycharmProjects/pythonProject1/main.py", line 4, in <module>
    torchvision.datasets.Caltech256("data", download=True)
  File "/Users/ikkamens/PycharmProjects/pythonProject1/venv/lib/python3.8/site-packages/torchvision/datasets/caltech.py", line 166, in __init__
    self.download()
  File "/Users/ikkamens/PycharmProjects/pythonProject1/venv/lib/python3.8/site-packages/torchvision/datasets/caltech.py", line 215, in download
    download_and_extract_archive(
  File "/Users/ikkamens/PycharmProjects/pythonProject1/venv/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 417, in download_and_extract_archive
    extract_archive(archive, extract_root, remove_finished)
  File "/Users/ikkamens/PycharmProjects/pythonProject1/venv/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 394, in extract_archive
    extractor(from_path, to_path, compression)
  File "/Users/ikkamens/PycharmProjects/pythonProject1/venv/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 260, in _extract_tar
    with tarfile.open(from_path, f"r:{compression[1:]}" if compression else "r") as tar:
  File "/Users/ikkamens/.pyenv/versions/3.8.2/lib/python3.8/tarfile.py", line 1604, in open
    raise ReadError("file could not be opened successfully")
tarfile.ReadError: file could not be opened successfully
Extracting data/caltech256/256_ObjectCategories.tar to data/caltech256

To Reproduce

Run:

import torchvision

torchvision.datasets.Caltech256("data", download=True)

Expected behavior

No exception, some images in the dataset in usable format.

Environment

Collecting environment information...
PyTorch version: 1.9.0
Is debug build: False
CUDA used to build PyTorch: None
ROCM used to build PyTorch: N/A

OS: macOS 10.14.6 (x86_64)
GCC version: Could not collect
Clang version: 11.0.0 (clang-1100.0.33.17)
CMake version: version 3.16.5
Libc version: N/A

Python version: 3.8 (64-bit runtime)
Python platform: macOS-10.14.6-x86_64-i386-64bit
Is CUDA available: False
CUDA runtime version: No CUDA
GPU models and configuration: No CUDA
Nvidia driver version: No CUDA
cuDNN version: No CUDA
HIP runtime version: N/A
MIOpen runtime version: N/A

Versions of relevant libraries:
[pip3] numpy==1.19.5
[pip3] torch==1.9.0
[pip3] torchvision==0.10.0
[conda] Could not collect

Additional context

A .tar archive is actually (partially?) downloaded. However, I can't extract it manually as well:
(venv) ~/P/p/c/caltech256 $ tar -xf caltech256/caltech256/256_ObjectCategories.tar imgs
tar: Error opening archive: Failed to open 'caltech256/caltech256/256_ObjectCategories.tar'

cc @fmassa @vfdev-5 @pmeier

[ikamensh]

p.s. similar thing happens for CalTech101:

/Users/ikkamens/PycharmProjects/pythonProject1/venv/bin/python /Users/ikkamens/PycharmProjects/pythonProject1/main.py
0it [00:00, ?it/s]
Traceback (most recent call last):
  File "/Users/ikkamens/.pyenv/versions/3.8.2/lib/python3.8/tarfile.py", line 2311, in next
    tarinfo = self.tarinfo.fromtarfile(self)
  File "/Users/ikkamens/.pyenv/versions/3.8.2/lib/python3.8/tarfile.py", line 1103, in fromtarfile
    obj = cls.frombuf(buf, tarfile.encoding, tarfile.errors)
  File "/Users/ikkamens/.pyenv/versions/3.8.2/lib/python3.8/tarfile.py", line 1039, in frombuf
    raise EmptyHeaderError("empty header")
tarfile.EmptyHeaderError: empty header

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/ikkamens/PycharmProjects/pythonProject1/main.py", line 4, in <module>
    torchvision.datasets.Caltech101("data", download=True)
  File "/Users/ikkamens/PycharmProjects/pythonProject1/venv/lib/python3.8/site-packages/torchvision/datasets/caltech.py", line 51, in __init__
    self.download()
  File "/Users/ikkamens/PycharmProjects/pythonProject1/venv/lib/python3.8/site-packages/torchvision/datasets/caltech.py", line 123, in download
    download_and_extract_archive(
  File "/Users/ikkamens/PycharmProjects/pythonProject1/venv/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 417, in download_and_extract_archive
    extract_archive(archive, extract_root, remove_finished)
  File "/Users/ikkamens/PycharmProjects/pythonProject1/venv/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 394, in extract_archive
    extractor(from_path, to_path, compression)
  File "/Users/ikkamens/PycharmProjects/pythonProject1/venv/lib/python3.8/site-packages/torchvision/datasets/utils.py", line 260, in _extract_tar
    with tarfile.open(from_path, f"r:{compression[1:]}" if compression else "r") as tar:
  File "/Users/ikkamens/.pyenv/versions/3.8.2/lib/python3.8/tarfile.py", line 1617, in open
    return func(name, filemode, fileobj, **kwargs)
  File "/Users/ikkamens/.pyenv/versions/3.8.2/lib/python3.8/tarfile.py", line 1670, in gzopen
    t = cls.taropen(name, mode, fileobj, **kwargs)
  File "/Users/ikkamens/.pyenv/versions/3.8.2/lib/python3.8/tarfile.py", line 1647, in taropen
    return cls(name, mode, fileobj, **kwargs)
  File "/Users/ikkamens/.pyenv/versions/3.8.2/lib/python3.8/tarfile.py", line 1510, in __init__
    self.firstmember = self.next()
  File "/Users/ikkamens/.pyenv/versions/3.8.2/lib/python3.8/tarfile.py", line 2326, in next
    raise ReadError("empty file")
tarfile.ReadError: empty file
Extracting data/caltech101/101_ObjectCategories.tar.gz to data/caltech101

[pmeier]

Duplicate of #4108, which was fixed in #4109.